Wireless localization has become an important issue in internet of things. However, its performance degrades significantly when there are only non-line-of-sight (NLOS) propagation paths instead of line-of-sight (LOS) propagation paths. Therefore, this paper proposes a novel localization algorithm to improve the localization accuracy, where we assume that the NLOS error satisfies the requirement of distance dependent model (DDM). In order to obtain a stronger NLOS tolerance than conventional methods, the proposed algorithm analyzes the localization geometry and employs a new localization parameter, namely the time product of arrival (TPOA), to construct the linear localization equation group (LLEQ). Meanwhile, a simple least-squares (LS) principle is applied to solve this LLEQ, resulting in a low complexity. We verify our algorithm by computer simulations, and the result demonstrates that the proposed algorithm yields a better performance than traditional algorithms in the NLOS-DDM environment.Index Terms-Distance dependent model, non-line-of-sight error, time product of arrival, wireless localization.